The Susquehanna River Basin, spanning over 70,000 km^2 across three states, is a critical source of water, transportation, and recreation in the mid-Atlantic region. The river also has a history of flooding, which can devastate communities along its path. During the summer, heavy rainfall associated with the remnants of tropical cyclones is a common trigger for flooding, while in the winter so-called rain-on-snow flooding, in which rain melts an existing heavy snowpack, is the most significant driver of heavy flood events. We used four datasets, the Japanese Meteorological Agency’s JRA-55 reanalysis; the L15 model developed by Livneh; the North American Land Data Assimilation System (NLDAS) with VIC land-surface model; and a version of the Department of Energy’s Energy Exascale Earth System Model (E3SM) to assess rain-on-snow flooding in the Susquehanna River basin from 1985 to 2005. Model fields for precipitation, runoff, and snow water equivalent were compiled into netcdf files with equal temporal resolution and identical units. These files were then analyzed to find rain-on-snow “events” associated with observed wintertime flooding, the criteria for which were determined empirically. We found that the JRA-55 dataset most realistically captures the magnitude of snowmelt and runoff associated with these events when compared with USGS streamgage observations. The L15 and NLDAS datasets display low biases in their representation of these quantities, while the E3SM dataset has high biases in runoff and snowmelt. This work emphasizes the need for careful consideration of dataset biases when attempting to identify historical flooding events.